Suzy libertarian-freely chooses A, given the choice between A and B, and does so for Reasons R. Had Suzy libertarian-freely chosen B, she would have done so for reasons R*. R and R* are distinct, and possibly incommensurable. Now, in the actual world, Suzy chose A, and her choice is supposed to be sufficiently explained by citing R. If somebody asks why Suzy chose A rather than B, we can simply cite R (along with the fact that A and B are mutually excluding).
The objector says: Alright, that Suzy libertarian-freely chose A seems to be sufficiently explained by citing R. She acted on R. However, why did she act on R? and why did she act on R rather than R*?
At first this sounds like a good question, but I’m not sure it is. This question treats Suzy acting on R ‘A-ly’ as though acting on R rather than R* were itself a choice. This is clearly wrong. For a reductio, suppose I give a reason R1 for why Suzy acted on R, (and that if she had acted on R* it would have been because of R1*) would you then just ask me why Suzy acted on R1, or, indeed, why she acted on R1 rather than R1*? This very quickly leads to an infinite regress of explanations. Such an infinite regress cannot exist. Notice this is stronger than ‘does not’ exist – if the question is only coherent if there can possibly be an answer, and if the only answer to it would be infinitely regressive, then if no question for which an answer must be infinitely regressive is coherent, the question cannot be coherent. One can try to get out of this by simply denying the PSR or something, but here they have to argue that the PSR is not even possibly true (and I doubt very much that they want to take on that task), or that infinitely regressive explanations are legitimate (which seems absurd at best, and certainly at least a higher price to pay than finding another solution).
So, something must be wrong with the question itself. I think there is something wrong with the question. One cannot meaningfully ask what reason Suzy had for acting on some of her reasons rather than others. After all, the reasons in R are supposed to include an exhaustive list of reasons for A, and thus any reason for R would be part of R. Therefore, it makes no sense to ask what reasons Suzy had for R. The infinite regress is stopped in its tracks.
I will leave aside, for the moment, the hard contrastive question of why Suzy did A rather than B, but I will note that if the preceding analysis is correct, there is no sensible question asking why (meaning, for what reason) Suzy acted on R rather than R*. Even the contrastive question has to be quarantined.